2008 servers freezing. 2003 ok though!

[joe90bass]

Edit: Typo on topic! I haven't time travelled and seen MSs future offering! - Fixed (Ric_)

I've been trying to bring 4 of our 5 Fujitsu-Siemens RX300s online with 2008 std 32bit server. Unfortunately they all randomnly freeze. Our existing 2003 SP2 servers are all fine. Fujitsu have taken one back to bench test and no surprise it is 100% fine for them!

- 3 servers are running 2008 with TS 32bit installed and Office 2007 enterprise(will be running 100 Thin Client)
- 1 server is running 2008 as a DC (demoted before sending this server to Fujitsu for testing), Backup Exec 12d and is destined to run all our apps
- 1 server (not online yet) will be running Exchange 2007

Our event logs show nothing unusual, the fujistu tools show nothing either. The servers are not under load, at the moment we only have 29 Thin client online. And the apps server does naff all as there's no point when it freezes!

As all is well with the server at Fujitsu HQ I can only assume it is something on my setup causing it, but surely if it was a config issue it would affect the 2003 servers?

I've thought about a pwer issue, but the new servers (and old file server) are plugged into APC UPSs which don't show that they're overloaded (old DC's are on own UPSs). And I think the Fujitsu tools would show if it was power.

I'm at a loss to this and it's now starting to cause us a problem as we need the Thin Client online, and our old Exchange server is out of disk space so the new one is desperately needed!

I'm open to all suggestions no matter how obvious they maybe to try!

HELP!

[SYNACK]

Heres a list of the stuff that I would check, you've probably gone through all of them anyhow but it may end up helping.

• Upgraded firmware - BIOS, RAID, Network etc
• Latest Drivers
• Full Memory test, several passes
• Disable overhead processes to isolate them ie virus/malware scans
• Look through managment log
• check grounding of rack if they are rack mounted
• run a faulting server with different version of windows and test to isolate
• switch on kernal debugging and pipe it to a remote workstation - this one should give you the best chance of finding it but is time consuming, Win 2k3 instructions here, couldn't find 2k8 instructions from MS but these ones from hp look like they may yeild the same results for 2k8.
• Write a script that writes the amount of non paged pool memory to disk or a network location every 20 seconds along with system temps
  

[joe90bass]

- Have upgraded BIOS and RAID firmware.
- Will look into drivers (I'll ask Fujitsu for the latest ones when I speak to them later)
- I'm going to try one with 2008 64bit tomorrow and may try 2003 on one of the others (would prefer 2008 for TS reasons!)
- Kernel debugging sounds good, I'll give that a go!

Thanks for ideas, am open to anything at the moment! The Fujitsu techie has basically said that as it's not a hardware fault, it;s working fine in their lab and as I didn't get them with 2008 preinstalled there's only so much they can do, though to be fair he has gone away for another look at it.......

[joe90bass]

After I found out we have 2 free calls to MS with the schools agreement I logged a call with them. After quite a lengthy process to establish the problem and every possible piece of information from the logs, registry, etc, being uploaded to them, they came back with remove the broadcom management suite as the logs showed some issues with it and they had seen it before on other setups! So far so good, the big test will be when the kids are back after half term.

Keeping my fingers crossed!

[joe90bass]

After I found out we have 2 free calls to MS with the schools agreement I logged a call with them. After quite a lengthy process to establish the problem and every possible piece of information from the logs, registry, etc, being uploaded to them, they came back with remove the broadcom management suite as the logs showed some issues with it and they had seen it before on other setups! So far so good, the big test will be when the kids are back after half term.

Keeping my fingers crossed!

In case anyone else has this issue, it does seem to have been the broadcom suite. Not one has frozen since I was using the software to team and loadbalance, as you do! Fujitsu have mentioned that there's an updated version of the suite, but for the moment I'll manage without any thanks!!

[plexer]

Great I'm glad you were able to get it sorted with the help of MS and also for reporting back what the problem was for anyone in the future.

Ben

[gshaw]

I've heard bad things about Broadcom software before so rather than use it I bought Intel Pro cards instead, looks like whoever warned me off it was right

[joe90bass]

I've heard bad things about Broadcom software before so rather than use it I bought Intel Pro cards instead, looks like whoever warned me off it was right

Yep, you owe them a pint or two

The machines were certified 2008 ready, and came with the cards builtin...

Obviously R&D, testing, fault finding, etc doesn't include plugging into a network and loadbalancing/fault tolerant setting the NICs..............